tokenmaxxing at – Some startups say higher AI token spending boosts productivity and “AI literacy,” while others warn it’s an unstable trend driven more by incentives than strategy. Misryoum reports.
A new tech obsession is spreading through American startups: spending more on AI “tokens” than last quarter.
The token spending race, and why engineers buy in
Kavitta Ghai, 29, co-founded Nectir with a simple internal target: engineers should use AI coding tools enough to matter—not just enough to try them. What began as minimum weekly token spend for Claude Code eventually became monthly expectations for each engineer, and Ghai says it worked.
Her company’s senior engineers. initially skeptical of AI coding tools. now describe them as an “army of coders.” That shift mirrors a broader change across the industry. where tokens have become a practical stand-in for how much compute an engineer relies on during day-to-day work.. In plain terms, the more “tokens” used, the more heavily the tools are employed—and the faster teams can iterate.
For startups, the appeal is obvious.. Coding assistants can turn “blank page” time into a series of drafts, refactors, and tests.. But what makes token spending notable isn’t the technology; it’s the culture around it.. Instead of treating AI tools as optional aids, some leaders treat them like infrastructure—and they measure adoption accordingly.
Startup leaders call it “force multiplier” behavior
Aron Solberg. 44. co-founded Risotto and calls the approach a “force multiplier.” His team uses AI models from multiple providers. and he describes token spending as something that scales small teams without requiring an immediate hiring sprint.. Risotto’s token budget. he says. has grown substantially over the past six months. reflecting a willingness to spend early to learn fast.
That mindset—spend now to accelerate—isn’t new in business. but tokens give it a sharper. more trackable form than many traditional investments.. Solberg frames it the way many founders do when resources are tight: if you want output. you have to fund the inputs.. In venture-backed environments. where speed can determine whether a product learns quickly enough to survive. “tokenmaxxing” becomes less a buzzword and more a lever.
Other startups describe similar internal logic.. Quang Hoang of Vybe says the company has considered minimum quotas and leans on policies that keep spending flexible during early build phases.. In these situations. the line between “encouraging usage” and “pushing consumption” can blur—especially when leaders believe the team is still learning how to build effectively with AI.
Incentives and accelerators turn token use into momentum
There’s also a market mechanism behind the trend: investors and accelerators often structure incentives around usage. Founders told Misryoum that token budgets can be treated as part of how a team is funded, not unlike allocating funds for cloud compute or developer tools.
Accelerators, in particular, can make it easier for startups to move quickly.. When credits are offered as part of participation, “spend less” stops being a tempting principle and becomes an unlikely strategy.. In at least one case described to Misryoum. startup teams used credits aggressively to access top models without the same immediate cost pressure.
The result is a feedback loop.. Engineers experiment more when friction is lower.. Products improve when iteration speeds up.. And when early gains show up—like faster personal productivity or smoother team workflows—leaders often treat high token usage as proof. even if it doesn’t fully explain which parts of the workflow were truly accelerated.
The pushback: costs, proxies, and the risk of burning the budget
Not every founder buys into tokenmaxxing as a philosophy. Rishabh Sambare, 23, co-founded Gale and wants to spend more, but he dislikes certain pricing structures for alternative tools. For him, stable subscriptions are a workaround that lets his team keep moving without volatile usage costs.
Others go further and question the underlying idea.. Weave’s founding engineer. Brennan Lupyrypa. calls tokenmaxxing “extremely stupid. ” arguing that incentives based on spending can turn into a proxy problem.. A team might chase token consumption rather than focusing on measurable product outcomes—bugs fixed, features delivered, customer value proven.
In Weave’s case. the company set up alerts when monthly token spending crosses a threshold. and Lupyrypa says many engineers hit that cap quickly.. But he still differentiates between “not strangling engineers” and rewarding the behavior itself.. Misryoum reported that for him, the danger isn’t AI usage; it’s turning usage metrics into the goal.
Why some founders expect the trend to fade
Brennan’s prediction is that tokenmaxxing won’t survive long once budgets tighten and CFOs demand clearer returns. That view reflects a common arc in tech: early experimentation often looks like overspending until the business is forced to translate activity into economics.
Misryoum also heard a competing argument from founders who still defend heavy token usage.. One founder described it as sensible when product-market fit is close enough that intense iteration could convert effort into growth.. In that framework. the question isn’t whether token spending feels excessive—it’s whether the spending compresses the timeline to results.
The tension here is likely to persist because the incentives are real on both sides.. Engineers want tools that help them ship.. Investors and accelerators want momentum.. CFOs want sustainability.. And startups. especially those scaling teams. must decide whether high token usage is a temporary learning phase or a permanent cost structure.
What “tokenmaxxing” means for American startup culture next
Tokenmaxxing, for better or worse, is reshaping how American startups think about productivity metrics.. It turns an invisible resource—AI compute—into something leaders can track, compare, and push.. Misryoum sees it as part of a broader cultural shift in tech management: when systems can be measured. people look for ways to manage them aggressively.
The practical question for startups is whether they can keep the benefits without the burn.. That likely means setting goals that tie AI usage to outcomes. not just consumption: fewer regressions. higher-quality code. faster customer feedback loops.. If token spending can be connected to those outcomes. the approach may evolve into disciplined AI operations rather than a fad.
For now. the industry remains split between founders treating tokens as a growth tool and founders treating the spending race as a risky distraction.. Either way. the direction is clear: AI tools are becoming central to how startups work—and the debate is increasingly about how much control. cost. and measurement teams should apply to that dependency.